The Multi-Period Conundrum of Private Market Performance Metrics

By Massimiliano Saccone, CFA, Founder & CEO, XTAL Strategies

 

One of the useful legacies of the now-vacated SEC’s Private Fund Advisers Rule[i] is a clear and current definition of the purpose of reported performance. Performance metrics should allow investors:

“…to compare private fund investments and comprehensively understand their existing investments and determine what to do holistically with their overall investment portfolio.”

Currently, adopted metrics do not seem to properly achieve the stated objective, as they do not allow generalized and unbiased comparability across asset classes or within the same private market cluster. To date, the private market industry has adopted approximations based on available research on this topic.

Performance Comparability: Fundamental Constraints

CFA Institute’s performance evaluation general principle[ii] is that “in the investment management industry, the time-weighted rate of return is the preferred performance measure.” According to CFA Institute’s rules, the use of time-weighted measures is subject to stringent conditions that are not compatible with the interim cash flow events of private fund investments: 

“…to usefully compare returns among alternate portfolios, either the pattern of cash flows must be the same for all the portfolios or the return measurement must be insensitive to cash flows.”

The recommended alternative in the case of portfolio “external cash flow” events, generally defined as capital that enters or exits a portfolio, is the use of money-weighted performance metrics, such as the internal rate of return (IRR) or the various multiples on investment capital (MOIC or TVPI), as a proxy of compounded returns. 

“External cash flows occur, for example, as contributions by the plan sponsor or payments out of the portfolio to pension plan beneficiaries or as deposits and withdrawals by investors, in the case of a mutual fund or a personal portfolio.”

It is worth noting that, with private fund investments, “deposits and withdrawals” are neither possible nor in the control of clients, i.e. investors. Contributions and distributions are decisions made by the portfolio manager and, as such, passively received by clients. Consequently, while IRR may be an internal yardstick for General Partners, it is not as informative for the Limited Partner in the context of the overall portfolio.

In table 1 below, a simplified comparison among three funds of the same 2008 vintage (with an identical commitment of 10 dollars, made on Dec. 31st, 2007) shows that different amounts of drawn capital (Drawn % column) and cash flows, happening with different timing (contributions with negative readings in parentheses and distributions with positive readings), can lead to identical IRR, DPI, and TVPI results.

Table 1: Comparison among buyout funds’ cashflow patterns and performance metrics. Source: XTAL for illustrative purposes only.
 

The Multi-Period Performance Measurement Challenge

The real issue for time-weighted metrics is the fact that portfolio performance is necessarily framed in a multi-period, since-inception perspective. 

A MOIC of 1.6x that takes 6 years to realize implies a different return than one that takes 3 years. A 20% IRR does not necessarily produce 20 cents on the dollar every year as an annualized compound average growth rate (the time-weighted reference of real-life wealth creation) implies. 

Money-weighted metrics like IRR and MOIC give up, by construction, any reference to time. While seemingly spot measures, IRR and MOIC are atemporal metrics. Therefore, they fail to specify when the performance they measure starts and ends and how much capital is being measured.

In spite of this evidence, a multi-period, since-inception return generalization for money-weighted measures is an accepted norm.

In the presence of interim cash flows, the commonly accepted rules indicate that time-weighted returns are approximated “proxy-compounding,” money-weighted returns: 

“Time-weighted return can be approximated by valuing the portfolio periodically, calculating a money-weighted return for each period between valuations and linking the results.”

Both IRR and so-called Dietz time-weighted returns, calculated for the fund’s sub-periods, are typically used for the linked approximations, but the substance does not change. 

“The modified Dietz formula is, in fact, a first-order approximation of the internal rate of return and is, therefore, a money- weighted return calculation for a single sub-period.”

Compounding approximated money-weighted measures (as proxies of average returns over given time horizons) generates since-inception time series that are heavily distorted by the implicit reinvestment assumptions. The longer the time horizon, the bigger the distortion. IRR is not an annualized compound average growth rate; a 20% IRR does not compound 20 cents on the dollar every year since inception (as time-weighting conventions imply).

Bias from Multi-Period, Money-Weighting Approximations

The misleading consequences of applying the compounding mechanism on since-inception, money-weighted returns (even under the label of Modified Dietz TWR) are visible in every time-series of private market returns, calculated in currently available PE indices, or aggregated horizon IRR performance.

Data from most PE index providers, summarized in a comprehensive research report[iii], shows 10-year horizon IRRs in the 14.3%-18.5% range – which implies that every fund underlying the analyzed universe has returned, on average, between approximately 3.8x and 5.45x the capital invested every 10 years.

Data sourced from another leading information provider[iv] shows (more realistically) the average TVPI multiple at 1.7x (for the vintages between 1996 and 2015 of the coherent clusters of funds) and indicates a substantial disconnect with the representation of cumulated performance over time, obtained by compounding since-inception IRRs.

Performance benchmarks and indices are a critical instrument accompanying the adoption, and assisting the portfolio and risk management, of private fund investments. If they are based on since-inception IRR (or modified Dietz TWR) metrics, the flaws of which investors may not be equipped to dispute, they can be seriously misleading.

In other words, without properly calculated time-weighted returns, any quarterly reporting, quartile statistics, or time series of returns add only the appearance of consistency and informative value for investors in the illiquid private market fund category.

Traditional Benchmarking Metrics: The Spillover Effect

For these reasons, all benchmarking exercises based on IRR or multiples fail, not just on the basis of traditionally identified limitations, but more so on amplified, multi-period grounds. IRR calculations are single-asset, single-period exercises. And the measures introduced to overcome the benchmarking limitations of IRR, albeit in relative performance terms, do not actually change the situation. The various versions of public market equivalent (PME) or alpha or excess value measures remain single-asset measures.  

IRR is a well-known shortcut for net present value (NPV) calculations at the single project level. PME is a relative value variation of this theoretical exercise and is only possible on an ex-post basis. The alphas and the excess values are variations of IRR and PME. 

As all these metrics do not consistently allow measuring and comparing performance of different investments with respect to amount and timing of cash flows, their averages lack statistical robustness. Therefore, rather than relying on a consequently impossible-to-compute private market beta / average returns, PME, alpha, and excess value measures adopt the approximation of a listed market benchmark index for their calculation. This exposes their outcomes to the unpredictable and random influence of unrelated volatility, as no direct beta peg relation exists between the listed index and the unlisted portfolio.

A Possible Solution               

While this discussion can reach high levels of theoretical sophistication, the matter boils down to the practical question of whether a 15% IRR can pay a 7% pension annuity over a given time horizon and a given amount of capital. An “it depends” reply on IRR, and less so PME and alpha information, would not help much. A possible solution to this conundrum will be proposed in a follow-up post. Stay tuned. 

About the Contributor
 

Massimiliano Saccone is the Founder and CEO of XTAL Strategies and developed its patented DARC methodology. Beforehand, he was a Managing Director, ultimately Global Head of Multi-Alternatives Strategies, at AIG Investments, after stints at DWS, Deloitte, and KPMG. A CFA charterholder, qualified accountant, and auditor, he holds a Master's in International Finance from the University of Pavia, and a magna cum laude Master's in Business and Economics from La Sapienza University in Rome. He is an active volunteer at the CFA Institute, and a Lieutenant of the Reserve of the Guardia di Finanza.

Learn more about CAIA Association and how to become part of a professional network that is shaping the future of investing, by visiting https://caia.org/